粒子群优化算法的参数调整策略研究

Research on Parameter Adjustment Strategy of Particle Swarm Optimization Algorithm

在粒子群优化算法中,对参数的调整会直接影响算法的收敛性、精确性和稳定性。为此,采用线性和非线性2种调整惯性权重策略,设置了3种调整方案,探索惯性权重调整对算法的影响。以Griewank、Rosenbrock等5个函数作为基准测试函数的仿真结果表明:Rosenbrock函数和Rastrigin函数收敛率低,且平均最优解与最优点存在很大偏差;Rosenbrock函数和Rastrigin函数时,在典型的线性递减策略中性能较差;在处理多峰函数问题时,取大值可提升算法的性能。

Different parameter settings in particle swarm optimization have different effects on the performance of the algorithm.Parameter adjustment directly affects the convergence,accuracy and stability of the algorithm.In view of this,linear and nonlinear inertia weight adjustment strategies were adopted,and three adjustment schemes were set to explore the influence of inertia weight adjustment on the algorithm.Five functions such as Griewank and Rosenbrock were used as benchmark functions.Simulation results showed that the convergence rate of Rosenbrock function and Rastrigin function was low,with a big deviation between the average optimal solution and the optimal advantage.Rosenbrock function and Rastrigin function had poor performance in typical linear decrement strategy.When dealing with multi-peak function problems,the performance of the algorithm could be improved by taking a large value.